A Novel Molecular Determinant for cAMP-dependent Regulation of the Frog Heart Na+-Ca2+Exchanger*

Na+-Ca2+ exchanger is one of the major sarcolemmal Ca2+ transporters of cardiac myocytes. In frog ventricular myocytes the exchanger is regulated by isoproterenol via a β-adrenoreceptor/adenylate-cyclase/cAMPdependent signaling pathway providing a molecular mechanism for the relaxant effect of the hormone. Here, we report on the presence of a novel exon of 27-base pair insertion, which generates a nucleotide binding motif (P-loop) in the frog cardiac Na+-Ca2+ exchanger. To examine the functional role of this motif, we constructed a full-length frog heart Na+-Ca2+ exchanger cDNA (fNCX1a) containing this exon. The functional expression of fNCX1a in oocytes showed characteristic voltage dependence, divalent (Ni2+, Cd2+) inhibition, and sensitivity to cAMP in a manner similar to that of native exchanger in frog myocytes. In oocytes expressing the dog heart NCX1 or the frog mutant (ΔfNCX1a) lacking the 9-amino acid exon, cAMP failed to regulate Na+-dependent Ca2+ uptake. We suggest that this motif is responsible for the observed cAMP-dependent functional differences between the frog and the mammalian hearts.

Na ؉ -Ca 2؉ exchanger is one of the major sarcolemmal Ca 2؉ transporters of cardiac myocytes. In frog ventricular myocytes the exchanger is regulated by isoproterenol via a ␤-adrenoreceptor/adenylate-cyclase/cAMPdependent signaling pathway providing a molecular mechanism for the relaxant effect of the hormone. Here, we report on the presence of a novel exon of 27-base pair insertion, which generates a nucleotide binding motif (P-loop) in the frog cardiac Na ؉ -Ca 2؉ exchanger. To examine the functional role of this motif, we constructed a full-length frog heart Na ؉ -Ca 2؉ exchanger cDNA (fNCX1a) containing this exon. The functional expression of fNCX1a in oocytes showed characteristic voltage dependence, divalent (Ni 2؉ , Cd 2؉ ) inhibition, and sensitivity to cAMP in a manner similar to that of native exchanger in frog myocytes. In oocytes expressing the dog heart NCX1 or the frog mutant (⌬fNCX1a) lacking the 9-amino acid exon, cAMP failed to regulate Na ؉ -dependent Ca 2؉ uptake. We suggest that this motif is responsible for the observed cAMP-dependent functional differences between the frog and the mammalian hearts.
The sarcolemmal Na ϩ -Ca 2ϩ exchanger is one of the major Ca 2ϩ extrusion pathways of the heart muscle. In nonmammalian species, the exchanger, in addition, may serve as a major Ca 2ϩ influx pathway, as these hearts in general lack well developed intracellular Ca 2ϩ release pools (1)(2)(3).
In mammalian species, most, if not all, tissues contain a transcript of the Na ϩ -Ca 2ϩ exchanger (NCX1) gene (4 -13) which undergoes tissue-specific alternative splicing. Two additional genes are expressed in the brain (NCX2 and NCX3) and skeletal muscle (NCX2), but the primary structure of the exchangers remains highly conserved, especially within the 11 putative transmembrane domains. Relatively greater divergence has been found in the N-terminal regions and the large intracellular loop between transmembrane domains 5 and 6, where the high affinity Ca 2ϩ regulatory site is located (14). Even though a putative protein kinase A phosphorylation site has been also identified in the mammalian isoform (4), no functional evidence for the cAMP/protein kinase A-dependent phosphorylation of the exchanger has as yet been found. Recently, however, it has been shown that the Na ϩ -Ca 2ϩ exchanger in frog but not in mammalian ventricular myocytes is regulated by isoproterenol via the activation of a ␤-adrenoreceptor/adenylate-cyclase/cAMPdependent pathway (15). In this report we describe the functional expression of a recombinant cAMP-sensitive frog heart Na ϩ -Ca 2ϩ exchanger construct (fNCX1a) with a newly identified 9-amino acid exon which renders the molecule regulatable by cAMP.
Photometry-To measure Na ϩ -dependent Ca 2ϩ uptake both control and the frog clone (fNCX1a)-expressing oocytes were injected with (50 ng/oocyte) Ca 2ϩ -sensitive photoprotein, aequorin (dissolved in 1 mM EDTA buffer), 3 to 4 h before the measurements. Aequorin-injected oocytes were loaded with Na ϩ by incubating them for 30 min at room temperature in K ϩ /Ca 2ϩ -free Barth's solution containing in mM: NaCl, 88; NaHCO 3 , 2.4; MgSO 4 , 0.82; HEPES, 15; pH 7.4 with NaOH, supplemented with 30 M nystatin (Sigma, prepared as 10 mM stock in Me 2 SO) and then transferred to K ϩ /Ca 2ϩ -free Barth's without nystatin (18). Photon emission was measured from Na ϩ -loaded oocytes placed in glass scintillation vials at 6-s intervals in a liquid scintillation counter (model LS250, Beckman). First, the basal photon emission was estimated in a scintillation vial filled with 1 ml of conditioning K ϩ /Ca 2ϩfree Barth's solution. The oocytes were then quickly transferred to a K ϩ /Na ϩ -free Barth's test solution containing in mM: triethanolamine chloride, 90; CaCl 2 , 0.41; Ca(NO 3 ) 2 , 0.33; MgSO 4 , 0.82; HEPES, 15; pH 7.4 with triethanolamine hydroxide, and the measurements were repeated. To measure the Ni 2ϩ sensitivity of the [Na ϩ ] i -dependent Ca 2ϩ uptake, photon emission was measured in the K ϩ /Na ϩ -free Barth's solution supplemented with 5 mM NiCl 2 prior to the use of test solutions.
Screening of the Library and the Construction of the Full-length Frog Na ϩ -Ca 2ϩ Exchanger-The screening of Xenopus laevis heart cDNA library using a probe NcoI-MluI (nucleotides 269 -2609) fragment of the dog cDNA (gift of K. D. Philipson, Los Angeles, CA) yielded two clones, H3 and H6. Further screening of the library did not produce clones which encode the N terminus. The remaining portion was retrieved from a X. laevis genomic library (Stratagene) using a PCR 1 -derived fragment of H3 as a probe (primers: forward, 5Ј-gtggcagttactattgttcgtcgtgga-3Ј, reverse, 5Ј-gctctttctgggtttcgcctggctt-3Ј). A positive clone X9 was found to contain a nucleotide sequence equivalent to the 1.8kilobase pair mammalian exon 2. The full-length clone, fNCX1a (3.2 kilobase pairs) was then constructed from X9 and H6 DNAs in pGEM-T vector (Promega, Madison, WI), using a region encompassing the ATG start codon up to the BbsI (corresponding to amino acid residue F568) of X9, and the remainder from H6, including the entire coding sequence and a 400-base pair 3Ј-UTR (to the SpeI site).
Modifications of the Expression Construct-In order to replace the whole 3Ј-untranslated region of fNCX1a with that of Na ϩ -glucose cotransporter clone, pMJC424 (19), the coding region of fNCX1a was amplified by 10 PCR cycles using a sense primer that hybridizes at positions 1-20 of the coding region containing a SalI restriction site before the ATG initiation codon, and an antisense primer that hybridizes to the last 20 nucleotides of the coding region and contains additionally a MluI restriction site downstream to the stop codon. The amplification reaction was carried out in the presence of Pfu DNA polymerase (Stratagene) and 100 ng of fNCX1a plasmid. The PCR fragment was purified by agarose gel electrophoresis, digested with SalI and MluI, and ligated into pMJC424 (19) from which the coding region of the Na ϩ -glucose co-transporter had been removed by digestion with SalI and MluI. The resultant plasmid, pF1a/MC, was sequenced.
During subcloning of the fNCX1a coding and 3Ј-UTRs into transcription-competent vector pAGA (20) (a gift of L. Birnbaumer, UCLA, CA) in order to reduce the chances of point mutations due to PCR, only the 5Ј-terminal part (position Ϫ1 to 252) of the coding sequence of fNCX1a was amplified by 10 PCR cycles. Amplification was performed using a sense primer (5Ј-atcaggtctcccATGGTTGTCCTTCTGCT-3Ј) that hybridized with the first 17 nucleotides of the coding region and contained a BsaI restriction site (underlined), which after digestion with BsaI gave a NcoI-compatible end, one nucleotide upstream of the initiation ATG codon. Antisense primer that hybridized with the few nucleotides downstream the MfeI site (position 252 of the coding region). The amplification reaction was carried out in the presence of Pfu DNA polymerase (Stratagene) and 100 ng of fNCX1a plasmid. To provide perfect initiation consensus sequence (27), the C at position Ϫ3 of the original fNCX1a was replaced with G. After purification by agarose gel electrophoresis, the PCR fragment was digested with BsaI and MfeI and ligated with MfeI (252 position of the coding region), SpeI (polylinker of pGEM-T plasmid) restriction fragment of fNCX1a into plasmid pAGA (20) between the NcoI and XbaI sites. The resultant plasmid was designated pF1a/AGA.
In Vitro Transcription-cRNA for the injection into the oocytes was synthesized in vitro from either plasmid pF1a/MC or pF1a/AGA using the mCAP TM mRNA capping kit from Stratagene according to manufacture's instructions. Each oocyte was injected with 50 nl of cRNA solution in water at concentration 0.1 mg/ml. Mutagenesis-Deletion of the 27-nucleotide fragment (positions 1913-1939) of fNCX1a cDNA clone was made according to Kunkel et al. (21). CJ236 cells (Invitrogen) were transformed with the pF1a/MC plasmid. 2ϫ YT medium supplemented with 0.25 g/ml uridine was inoculated with a colony of recombinant clone, and the single-stranded form of the plasmid was isolated using R408 helper phage (Stratagene). A 34-base oligonucleotide complementary to 18 nucleotides upstream and 16 nucleotides downstream of the target sequence was phosphorylated with ATP and T4 polynucleotide kinase and annealed to the single-stranded plasmid. Conversion into closed circular doublestranded form was made with T4 DNA polymerase and T4 DNA ligase. The double-stranded plasmid was used for transformation of XL1-Blue cells (Stratagene). To confirm deletion, plasmid DNAs from several recombinant colonies were sequenced, and the clone containing the desired deletion was selected and designated pF1a/MC(⌬).

A Nucleotide-binding Motif in a Frog Na ϩ -Ca 2ϩ Exchanger-
The X. laevis heart cDNA library was screened with the cDNA pTB11, encoding dog heart Na ϩ -Ca 2ϩ exchanger as a probe. Two overlapping partial cDNA clones, H3 (corresponding to amino acid residues 388 -656) and H6 (residue 544-3Ј-4TR), were isolated and sequenced (GenBank TM accession no. X90838). Unlike H3, the H6 clone contained a 27-base pair insert located at a splicing junction between exons 7 and 8 (22) and representing a new exon (30).
The deduced amino acid sequence of H6 and H3 showed 87.8% identity to the corresponding part of the dog sequence. However, unlike mammalian clones, the 27-base pair insertion generates an ATP/GTP binding motif (P-loop) by adding the GKS sequence to GKILY (essential amino acids are underlined), thus forming the consensus P-loop structure (A/G)-X 4 -G-K-(S/T) (25). In the sequence encoded by H3 and H6, the putative protein kinase A phosphorylation site (4), and the Ca 2ϩ -binding domain (26) seen in the mammalian NCX1 proteins are conserved (Fig. 1). To examine the functional role of this motif, we constructed the full-length frog Na ϩ -Ca 2ϩ exchanger. Since the frog heart cDNA library lacked coding sequences for the N terminus of the Na ϩ -Ca 2ϩ exchanger, the X. laevis genomic library was screened using a fragment (nucleotides 1626 -1773) produced by PCR at the N terminus of the H3 clone. The stretch coding for the remainder of the region was found in X9 (GenBank TM accession no. X90839). A small diversity was found in the region of the overlap between H3/H6 and X9 (534 nucleotides, 178 amino acids). Alignment of the amino acid sequences showed the identity of 92.1%, and the similarity of 94.5% (1-amino acid insertion, 13-amino acid changes, and 17-nucleotide substitutions which do not cause changes in amino acid residues). It should be noted that the portion that was supplemented by the genomic sequence does not contain any known regulatory motifs (4). We therefore constructed the full-length "composite" frog clone (fNCX1a) for expression studies in Xenopus oocytes.
Functional Expression of the Frog Na ϩ -Ca 2ϩ Exchanger-Initially we were unable to express fNCX1a in Xenopus oocytes. Two strategies were therefore employed to achieve functional expression. First strategy involved the replacement of the whole 3Ј-UTR of fNCX1a with that of Na ϩ -glucose co-transporter clone pMJC424 (14), which includes a poly(A) tail (19) (gift of E. Wright, UCLA, CA). In addition, we replaced G for C in the Ϫ3 position to conform to a Kozak consensus sequence (27) and designated it pF1a/MC. In the second strategy, we replaced the 5Ј-UTR of fNCX1a with that of alfalfa mosaic virus RNA-4 and attached a 90-nucleotide poly(A) tail to the 3Ј-UTR of fNCX1a. This was done by subcloning the fNCX1a coding and 3Ј-UTR into the transcription-competent vector pAGA (20) Ni 2ϩ (as indicated) in response to the ramp voltage-clamp protocol shown above the records. The Ni 2ϩ -sensitive I Na-Ca was obtained after subtraction of the current recorded in presence of Ni 2ϩ from those of control. E, current-voltage relation for Ni 2ϩ -sensitive I Na-Ca constructed from the ramp portion of the difference current presented in D. F, the differences in Na ϩ -dependent Ca 2ϩ uptake in Na ϩ -loaded control and fNCX1a-injected oocytes in the presence and absence of 5 mM Ni 2ϩ as estimated by photon emission of aequorin. Oocytes from both groups were injected with aequorin and loaded with Na ϩ by incubation for 30 min at room temperature in conditioning K ϩ /Ca 2ϩ -free Barth's solution supplemented with 30 M nystatin. Photon emission was measured in both conditioning solution and following exposure of the oocytes to the test, K ϩ /Na ϩ -free Barth's solution. The number of oocytes tested is indicated above each column. The asterisk (*) denotes significantly different values from the value in test, K ϩ /Na ϩ -free Barth's solution at p Ͻ 0.05.
(a gift of L. Birnbaumer, UCLA, CA). This plasmid was designated pF1a/AGA. We have chosen this strategy to verify: (a) if the whole 3Ј-UTR or only the poly(A) tail were important for translation, and (b) if the efficiency of the translation could be further increased by the presence of the 5Ј-UTR of alfalfa mosaic virus RNA-4. We found that the pF1a/AGA construct produced a 50% higher level of expression of the exchanger than did pF1a/MC (using I Na-Ca in oocytes as a criterion). However, expression of pF1a/AGA was transient, reaching peak values within 24 -48 h, and declining rapidly, while pF1a/MC produced a slow but steady level of expression for up to 6 -7 days. Thus, the right Kozak consensus initiation site at the 5Ј-end and the presence of the poly(A) tail at the 3Ј-UTR are critical for the functional expression of the exchanger. Most of the experiments reported here were carried out using the pF1a/MC construct.
The frog exchanger cRNA was synthesized by transcription in vitro of the modified fNCX1a cDNA. Functional expression of the exchanger molecule was assessed 2-4 days later by direct measurements of the expressed current or monitoring of Na ϩdependent Ca 2ϩ uptake. Electrophysiological experiments were performed in Cl Ϫ -free extra-and intracellular solutions, using the "glass-funnel" technique that permits both fast voltage-clamp and intracellular perfusion of devitellinated oocytes (16). Na ϩ -dependent Ca 2ϩ uptake was determined by measuring photon emission of Na ϩ -loaded oocytes injected with the Ca 2ϩ -sensitive photoprotein, aequorin, after exposure to Na ϩfree solution. Fig. 2 illustrates the procedures used to isolate the inward and outward components of the membrane current carried by the Na ϩ -Ca 2ϩ exchanger in fNCX1a-injected Xenopus oocytes. The standard Cl Ϫ -free experimental solutions contained 109 mM Na ϩ and 5 mM Ca 2ϩ on the outside and 20 mM Na ϩ and 0.1-10 M Ca 2ϩ on the inside. Concentrations of free Ca 2ϩ in the intracellular solution were estimated according to the Ca-Buf program (SPECS) (28). In Cl Ϫ -free internal and external solutions depolarization of the oocyte from a holding potential of Ϫ60 to ϩ40 mV activated an outward current which was followed by a tail current on repolarization to Ϫ80 mV ( Fig. 2A,  lower panel). As the duration of the clamp pulse was prolonged, the outward current decayed slowly, and the tail currents following repolarization to Ϫ80 mV were enhanced ( Fig. 2A). Exposure of the myocytes to 3 mM Ni 2ϩ blocked a significant portion of both outward and accompanying inward tail currents (Fig. 2B). Subtraction of currents obtained, in the presence and absence of Ni 2ϩ , yielded a slowly decaying outward current followed by an expanding inward tail envelope (Fig. 2C), representing, respectively, the Ca 2ϩ influx and Ca 2ϩ efflux modes of the exchanger (15). 1 mM Cd 2ϩ similarly suppressed the current generated by the Na ϩ -Ca 2ϩ exchanger in fNCX1ainjected oocytes (data not shown). Fig. 2F documents the results of a series of experiments showing the differences in Na ϩ -dependent Ca 2ϩ uptake in the Na ϩ -loaded control and fNCX1a-injected oocytes, and their sensitivity to Ni 2ϩ . Photoluminescence in oocytes (Ca 2ϩ uptake) was measured in both conditioning control and K ϩ /Na ϩfree Barth's test solutions. In control oocytes no differences in photoluminescence were found between the conditioning and test solutions (Fig. 2F). On the other hand, oocytes injected with fNCX1a showed approximately 15-fold higher photoluminescence in the test compared with conditioning solutions (Fig.  2F), suggesting significant Na ϩ -dependent Ca 2ϩ uptake in response to the expression of fNCX1a. Photoluminescence in fNCX1a-injected oocytes was almost completely blocked by addition of 5 mM Ni 2ϩ (Fig. 2F).
To measure voltage-dependence of the exchanger, the oo-cytes were first depolarized to ϩ40 mV and then the voltage was linearly changed at 200 mV/s to Ϫ120 mV (ramp clamp protocol, Fig. 2D, upper panel). The lower panel of Fig. 2D shows superimposed traces of control currents, the current in the presence of 3 mM Ni 2ϩ , and the difference currents (Ni 2ϩsensitive I Na-Ca ) activated by such a pulse protocol. The Ni 2ϩsensitive I Na-Ca had a reversal potential (E Na-Ca ) at ϩ20 mV (Fig. 2E). The average experimental value for E Na-Ca in 14 oocytes from different frogs injected with different samples of  Fig. 2, A-D) might accumulate in a confined intracellular space in the vicinity of the membrane.
To verify this hypothesis we employed a pulse protocol in which the negative voltage ramp was preceded by progressively longer conditioning depolarization to ϩ40 mV, thus increasing the entry of Ca 2ϩ into the oocyte prior to the application of voltage ramp (Fig. 3A, upper panel). Fig. 3A (lower panel) shows that prolonging the depolarizing pulse at 40 mV shifted the reversal potential of the exchanger current to more positive values of 3.2, 15, 21.8, 26.8, and 31.8 mV, respectively (Fig. 3B). At 5 mM [Ca 2ϩ ] o and with external and internal Na ϩ of 109 and 20 mM, respectively, the measured reversal potential suggests an effective [Ca 2ϩ ] i of 35, 55.5, 72.3, 88, and 107 M, using E Na-Ca ϭ 3E Na Ϫ 2E Ca . The experimentally measured shifts of the reversal potential, ⌬E Na-Ca ϭ E Na-Ca(1) Ϫ E Na-Ca(2) , corresponded well with the changes in the effective internal Ca 2ϩ concentration in accordance with the relation, ⌬E Na-Ca ϭ (2) ), supporting the idea that Ca 2ϩ entering the oocyte via the exchanger accumulates in a confined intracellular space in the proximity of the membrane.
Digital integration of the current traces of Fig. 3A provided an estimate of the total amount of Ca 2ϩ entering the oocyte until the exchanger reaches its reversal potential. The total amount of Ca 2ϩ transported and the effective [Ca 2ϩ ] i , determined from the measurements of the reversal potential, allows estimation of the size of the space equivalent to 27 nl, i.e. ϳ2.7% of the average volume of the oocyte (1 l).
Modulation of Exchanger Activity by cAMP-In the frog ventricular myocytes the Na ϩ -Ca 2ϩ exchanger is suppressed via the ␤-adrenoreceptor/adenylate-cyclase/cAMP-dependent pathway (15). Since the defolliculated and devitellinated oocytes lack ␤-adrenoreceptorand forskolin-stimulated adenylate cyclase (29), the membrane-permeable cAMP analog CPT-cAMP was used to determine the cAMP sensitivity of fNCX1a-injected oocytes. Fig. 4A shows that I Na-Ca at 40 mV is slowly suppressed by 50 -60% (n ϭ 4) on rapid (Ͻ500 ms) application of CPT-cAMP. In the presence of CPT-cAMP, 5 mM Ni 2ϩ failed to further inhibit the current, suggesting that virtually all the Ni 2ϩ -sensitive I Na-Ca had been inhibited by CPT-cAMP. The CPT-cAMP-mediated suppression of the exchanger current was partially reversed by the wash out of extracellular CPT-cAMP, and the extensive perfusion of the oocyte's interior with cAMPfree internal solution (Fig. 4A). The recovery of the net current was primarily due to the recovery of I Na-Ca , since all of the recovered current was blocked by 5 mM Ni 2ϩ (Fig. 4A). Fig. 4, B and C, shows the CPT-cAMP effect on I Na-Ca recorded in another fNCX1a-injected oocyte, using the envelopepulse protocol (Fig. 4B, upper panel). In this oocyte, 6-min exposure to CPT-cAMP resulted in 50% inhibition of the current measured at both the holding potential (Ϫ60 mV) and during depolarization to 40 mV. The tail currents accompanying membrane repolarization to Ϫ80 mV were virtually abolished in CPT-cAMP-containing solutions, suggesting a strong suppressive effect of cAMP on Ca 2ϩ efflux as well as Ca 2ϩ influx modes of the exchanger (n ϭ 4).
The Role of the 9-Amino Acid Motif in cAMP-dependent Modulation of Na ϩ -Ca 2ϩ Exchanger-Since the presence of the 9-amino acid insertion comprising a putative nucleotide binding domain is the major distinguishing structural feature of the frog cardiac Na ϩ -Ca 2ϩ exchanger (fNCX1a), we constructed a deletion mutant, ⌬fNCX1a, without the 9-amino acid insertion. Fig. 5 compares the differences in Na ϩ -dependent Ca 2ϩ uptake measured by photoemission of aequorin in oocytes injected with the dog heart NCX1, the frog heart fNCX1a, and the mutated frog heart ⌬fNCX1a exchangers. Oocytes from each group were separated into two subgroups, one of which following the Na ϩ loading period was maintained in control conditioning solution and exposed to the control test solution, while the other was maintained in conditioning solution and subjected to the test solutions supplemented with 500 M CPT-cAMP. The histograms of Fig. 5 show no significant differences in Na ϩ -dependent Ca 2ϩ uptake between the control and CPT-cAMP exposed oocytes expressing either the dog heart NCX1 or the mutated frog heart ⌬fNCX1a exchangers, both of which lack the 9-amino acid nucleotide binding domain. In sharp contrast, the Na ϩ -dependent Ca 2ϩ uptake was significantly smaller in the presence of CPT-cAMP in the oocytes expressing the frog heart fNCX1a isoform, consistent with the idea that the 9-amino acid domain, is critical for cAMP-mediated regulation of the exchanger. DISCUSSION We have successfully expressed a functional cAMP-regulated frog Na ϩ -Ca 2ϩ exchanger (fNCX1a) in Xenopus oocytes. The most distinguishing structural feature of the construct is the presence of a consensus ATP/GTP binding 9-amino acid motif (P-loop), located between residues 636 and 646 of the main cytoplasmic linker. Deletion of this motif from the clone (⌬fNCX1a) abolished the cAMP-dependent regulation of the exchanger (Fig. 5). fNCX1a, constructed from H3/H6 and a genomic clone X9, contained a small (7.9%) divergence in the overlapping 178-amino acid sequence and only 6.4% divergence at the nucleotide level at the same region. This divergence is in the order of interspecies polymorphism for orthologous genes, since it is substantially lower than that observed between the genes of the same family in mammals (e.g. 35-40% divergence in rat) (23). Whether this divergence arises from polymorphic variation of the frog subspecies (where the structure of genus Xenopus is not well defined) or from the unlikely presence of a novel member of NCX family is not as yet clear. Irrespective of whether the clone represents a chimera of orthologous genes from related species, the presence of the putative protein kinase A phosphorylation site and Ca 2ϩ binding domain, the similarity in cAMP-mediated regulation between the expressed protein and the exchanger in native frog myocytes (15), as well as the abundance of fNCX1a sequence in mRNA of frog heart compared with other tissues, 2 suggests that the clone represents a legitimate molecular model to study the functional implication of the novel motif.
Voltage clamp studies in early 1970s have shown that developed tension in frog heart has a dominant tonic and a small phasic component (1,2,31). In the presence of catecholamines, however, the phasic (I Ca -dependent) component of tension is strongly enhanced, while the sigmoid Na ϩ -dependent tonic component is strongly suppressed (32). This dual effect of catecholamines was thought to result from both increased Ca 2ϩ influx via the Ca 2ϩ channels and enhanced uptake of Ca 2ϩ by the sarcoplasmic reticulum. In light of recent studies showing the virtual absence of SERCA II gene of Ca-ATPase and phospholamban proteins in the frog heart (33-35), the exchanger may be the molecular site that mediates the relaxant effects of catecholamines. At first the suppressive effect of cAMP on the exchanger appears contraintuitive, but considering that the Ca 2ϩ channel and the plateau of the action potential are significantly enhanced in the presence of catecholamines (32,36), necessarily increasing Ca 2ϩ influx via the exchanger, the cAMPdependent suppression of the exchanger may be the appropriate evolutionary solution to stem the tide of large Ca 2ϩ influx that would result. Thus, the suppression of the tonic Ca 2ϩ influx pathway may contribute to the early fall in tension observed during depolarizing pulses in the presence of isoproternol (32).
The ␤-agonist/protein kinase A-induced temporal shift of the fraction of contractile Ca 2ϩ , transported into the cell during the action potential from the exchanger to the Ca 2ϩ channel, may be related to the evolutionary requirement of the fight and flight reflex in almost all nonmammalian vertebrates (and possibly prenatal mammals) lacking significant SERCA II and intracellular Ca 2ϩ release pools. Thus, the heart of these animals may utilize the same ␤-adrenergic regulatory mechanism for both the control of phasic (Ca 2ϩ channel) and tonic (exchanger) transport of Ca 2ϩ and development of tension. Under sedentary conditions the exchanger would primarily deliver 2 A. Kraev and E. Carafoli, unpublished data.
FIG. 5. The differences in Na ؉ -dependent Ca 2؉ uptake in Na ؉loaded oocytes injected with dog heart NCX1, intact frog heart fNCX1a, and mutated frog heart fNCX1a(⌬) Na ϩ -Ca 2ϩ exchangers in the presence and absence of 500 M CPT-cAMP. Oocytes from all groups were injected with aequorin and loaded with Na ϩ by incubation for 30 min at room temperature in conditioning K ϩ /Ca 2ϩfree Barth's solution supplemented with 30 M nystatin. After Na ϩ loading control subgroups of the oocytes were maintained in regular conditioning solution, whereas test subgroups of the oocytes were maintained in conditioning solution supplemented with 500 M of CPT-cAMP. Photon emission was measured following exposure of the oocytes from the control subgroups to the regular test, K ϩ /Na ϩ -free Barth's solution and oocytes from the test subgroups to the test, K ϩ /Na ϩ -free Barth's solution supplemented with 500 M CPT-cAMP. The number of oocytes tested is indicated above each column. The asterisk (*) denotes a significantly different value from the corresponding control value at p Ͻ 0.05. and extrude the contractile Ca 2ϩ into and out of the cell. Upon sympathetic stimulation, as the heart shifts to the faster Ca 2ϩ delivery pathway via the phosphorylated Ca 2ϩ channel, the influx of Ca 2ϩ via the exchanger would be suppressed, providing the heart with faster but shorter contractions to accommodate the faster heart rate. Whether such a protein kinase A-dependent-regulatory mechanism can be made to operate by genetic manipulation of the mammalian Na ϩ -Ca 2ϩ exchanger when the exchanger is overexpressed (37,38) in heart failure remains to be tested.